On Conlang Herrig Thaillí (Henrik Theilling) wrote in response to me:
I’m ATM in woes WRT the Rhodrese indefinite article. I feel that the changes I’ve made to the feminine indefinite and plural definite forms call for a change in the plural indefinite as well. Consider the following patterns:
masc. sing. fem. sing. plur. _#C _#V _#C _#V _#C _#V def. el el la l’ li gl’ indef. un un na n’ eun eun OR
masc. sing. fem. sing. plur. _#C _#V _#C _#V _#C _#V def. el el la l’ li gl’ indef. un un na n’ ni gn’ Is the latter preferable or am I over-regularizing?
NB eun would still mean ‘some, a few’, while aocú means ‘some, any’ and naocú means ‘not any, none’.
My gut feeling for Rhodese is that the first alternative is more like it. It has that nice vowel change. And the system should not be made too regular I think.
That’s exacly what one part of me is saying, but I’m not qute sure that the language doesn’t want otherwise…
Note that my answer disregards any aspect but aesthetics, because I basically have no idea how the modern words are derived exactly and why you would think you’re overgeneralising in the second alternative.
The basic idea is that since the forms of ille which ended in a long vowel developed forms stressed on the ending, why could not the forms of unus do the same? There are essentially two objections:
- unus probably became an indefinite article much later than ille became a definite article, and so should not develop in parallel with it.
- ūnus had a long vowel in the first syllable, which would be a stress attractor.
Contra objection 1. can be said that the field should be rife for analogies with the definite article.
Contra objection 2. can be said that ille also began with a heavy syllable, due to the double ll.
In fact I suspect that one factor which made the ending-stressed forms of ille arise in the first place may have been the way assignment of secondary stress affected them once they became proclitic. Recall that secondary stress in Latin apparently tended to fall on every second syllable before the main stress:
ˌin-cre-ˈdi-bi-lis
apparently a first syllable immediately before the stressed syllable received a secondary stress even though no unstressed syllable intervened:
ˌcre-ˈdi-bi-lis
but I think that a first syllable of a word with three syllables before the stressed syllable (which was uncommon in Latin) did not, or at least did not always receive secondary stress, or attract it from the following syllable:
pe-ˌrin-cre-ˈdi-bi-lis
At least that seems to be the pattern which seems to apply in modern Italian.
Now consider what would happen when the disyllabic forms of ille were procliticized to words stressed on the first and second syllable respectively:
ˌil-la-ˈta-bu-la
but
il-ˌla-ta-ˈber-na
The nominative singular masculine apparently got special treatment. The short unstressed final ĕ must have been prone to disappear for its own reasons as part of the cliticization process; probably it was already [ə] already and could not receive contextual stress. Whatever the particular reason we get
ˌil-ca-ˈbal-lus
as well as
ˌil-ˈpa-tre
(NB French le cheval, le père are from the accusative with illum — it was still lo in Old French.)
Now what I think happened nect was that the unstressed initial vowel in il-ˌla-ta-ˈber-na fell off, giving ˌla-ta-ˈber-na, and the thus arisen allomorph la was later generalized to all cases. Actually I think the burden of proof rests on those who would claim that forms of ille received stress on the endings in some other way!
Finally even if the ‘articulization’ of unus was later it shared two important features with ille: it was disyllabic and it began with a vowel, which in Vulgar Latin was just as short as i in ille in proclitic position. To be sure unus caballus would not regularly become un caballus but rather *nus caballus according to my theory above, but unus pater would become un(u)s pater and in the accusative unu patre with stress on un and pa, and analogy with the definite article could take care of the rest.
Note that I really miss the u in some of the indefinite articles. For me, u is the essence of that article, not n, but of course, that’s pure aesthetics again. :-) Maybe that’s why I like eun more than ni. (Terkunan has nus with an u…)
To me the essence of the Romance definite article is l, yet look at Portuguese and the Italian masculine plural! :-)
There would be some factors in Rhodrese which could tip the scales towards the n-based forms: the regular Old Rhodrese reflex of *uni before a pause or a fricative would be *eu, but that would be homophonous with habes ‘thou hast’, which probably would favor forms with preserved n — though that may just as well be eun, so that’s another matter. However there is one possibility which I imagine could have happened in some Romance natlang somewhere, though I never heard of it, namely that ūnus got reanalysed as ŭnnus on analogy with ille — that would both explain Rhodrese stressed forms with -n and encourage total analogy with ĭlle as per above. I guess compound forms like quisque-cata-unus and aliqui-unus could might be exempt from this and keep single *-n which would then be lost before a pause or a fricative as usual in Rhodrese1.
I need to sort out the possible rôle of pronominal eun. Your overview of Spanish indefinite pronouns will be most helpful! Now if I could only find that French grammar…
BTW the name of the language is Rhodrese with a dr in the middle — it derives from Rhódanus Rhuodre ‘Rhône’ and has nothing to do with Rhodes Rhodus Rhuod, although the names end up similar. I used to have a list of other words with d’n > dr somewhere. Primary dr of course becomes rr, as does nr.
-
I operate with the hopefully plausible assumption that before a stop there was free variation between
nasalized vowel + nasal + stopand
nasalized vowel + stopwhere the vowel nasalization would later get lost, which gives quite some room for variation in resulting forms in standard modern Rhodrese where word boundaries are involved. ↩